摘要
以变压器油中溶解气体含量的分析结果为基础，利用人工神经网络技术可比较有效地解决电力变压器的故障诊断问题。本文引入一个组合神经网络模型以实现对变压器绝缘故障的多分辨识别，并在此基础上结合多元统计分析技术初步实现了组合神经网络模型中训练样本集的典型性筛选，仿真结果也显示出所提方法的有效性。
Based on the neural nctwork modcl, internalfaults of the transformers could be diagnosed aocording tothe test data obtaincd frorn the dissolved gas analysis in oll.In this paper, a Combinatorial Neural Network is intro-duced to realize the rnulti-resolution rccognition to the insu-lation fault of power transformer. And the sclection oftraining samples could be rcalized on the basis of the multi-variate statistical analysis. The s imulation results show thatthis method is effective.
出处
《高电压技术》
EI
CAS
CSCD
北大核心
1999年第4期1-3,6,共4页
High Voltage Engineering
基金
国家自然科学基金
东北电力集团资助!59637200
关键词
电力变压器
故障诊断
神经网络
多元统计
power transformer fault diagnosis neural network multivariate statistics
